✕

About PictureFLAME

PictureFLAME animates FLAME routines, showing how operands (vectors and/or matrices) are repartitioned and updated step by step.

Create a routine

This is usually done by generating a code skeleton with Spark and then modifying the result.

Loading a program

Paste or type a FLAME routine (MATLAB/Octave or Python) in the left panel. Select the language from the dropdown, then click Load. PictureFLAME parses the code and builds initial operand vectors and matrices from the function/routine signature.

Animating

• Step — advance one logical FLAME step (a repartition or continuation group).
• Play / Stop — run continuously at the speed set by the Fast/Slow slider.
• New — reset to the initial state so you can replay from the beginning.

Verification formula

Type a mathematical expression in the Verification formula box (e.g. A * x + y) to check whether the output operand matches the formula upon completion. Operand names and basic matrix arithmetic (+, -, *, transpose T(x) or x') are supported.

Tips

• Highlighted cells in the operand display show which elements are being read or written at the current step.
• Operand sizes are inferred from the function signature — rename operands to match your own notation.

Acknowledgments

PictureFLAME was created by Robert van de Geijn (UT Austin) and Tze Meng Low (Carnegie Mellon University).

The latest rewrite of PictureFLAME was supported in part by the National Science Foundation (NSF) and the U.S. Department of Energy (DOE), Office of Science, Office of Advanced Scientific Computing Research, under the following awards:

• CCF-2446143 (CMU), CCF-2446145 (UT Austin), DE-SC0025950 (SMU) — Collaborative Research: CS2: Deriving Correct Programs for Performant Computational Chemistry, NSF program: Correctness in Scientific Computing Systems (CS2)
• OAC-2513927, OAC-2513928, OAC-2513929 — Collaborative Research: Frameworks: Framework for Advanced (Multi)Linear Infrastructure in Engineering and Science (FAMLIES), NSF program: Cyberinfrastructure for Sustained Scientific Innovation (CSSI)

Any opinions, findings, conclusions, or recommendations expressed here are those of the authors and do not necessarily reflect the views of the NSF or DOE.